Updated container has gone all wrong

[Femur4889]

I simply want to make my updated JF container behave exactly like my old established installation did.

I'm really at a loss with what to do, I've troubleshot this for hours but nothing works. 

I had not used Jellyfin in quite some time. I updated the docker container using docker compose. The container loaded but when I entered into Jellyfin on my server, none of the media files were there and my original Username/Password combo did not work.

I do not want to lose all the metadata/info which was associated with my old server.

I am certain that my media path is correct. My existing /config and /cache files are currently in the same place they've always been. My existing /media has been moved to a local ZFS array and I have changed the path's settings in the docker compose file.

version: '3.5'
services:
  jellyfin:
    image: jellyfin/jellyfin
    container_name: jellyfin
    #user: 1000:1000
    network_mode: 'host'
    volumes:
      - /mnt/500gb/docker/jellyfin/configconfig
      - /mnt/500gb/docker/jellyfin/cachecache
      - type: bind
        source: /mnt/gator1/Jellyfin/
        target: /media
        read_only: true
    restart: 'unless-stopped'
    extra_hosts:
      - 'host.docker.internal:host-gateway'

If anyone can assist I'd be most grateful.

[TheDreadPirate]

Besides upgrading to 10.8.13, did you also change from lsio/jellyfin to jellyfin/jellyfin? Do you have more than one jellyfin container running?

[TheDreadPirate]

Also, why is the user field commented out? With it commented out, the container is run as root. This should mean there aren't permission issues for file and folder accesses as long as the volume parameters are correct.

[Femur4889]

Also, why is the user field commented out?  With it commented out, the container is run as root.  This should mean there aren't permission issues for file and folder accesses as long as the volume parameters are correct.

Thanks for your reply. 

If I uncomment User then I am totally unable to load JF in the browser.

Yes, I only have one JF container running. If I run 

Code:

docker ps -a

i see only this JF container.

My original JF docker container was always 

Code:

jellyfin/jellyfin

and that is also the one I'm using currently.

If User is commented out then I am able to load JF in the browser, although from here none of my original settings or library are loaded. If from here I try to load media files, when I click "ok" a graphic just spins and spins.

[TheDreadPirate]

I forgot about this. You need to add "group_add" in your docker compose. Mine for reference.

Code:

services:
  jellyfin:
    image: jellyfin/jellyfin:unstable
    container_name: jellyfin-unstable
    user: 105:104
    group_add:
      - "993" # Intel render group
    network_mode: 'host'
    volumes:
      - ./jellyfin-data/config:/config
      - ./jellyfin-data/cache:/cache
      - /media/storage2:/media/storage2:ro
      - /media/library:/media/library:ro
    devices:
      - /dev/dri/renderD128:/dev/dri/renderD128
    restart: 'unless-stopped'
    extra_hosts:
      - "host.docker.internal:host-gateway"

Use the command "getent group render" and replace "993" in mine with your system's render group ID.

[Femur4889]

I forgot about this.  You need to add "group_add" in your docker compose.  Mine for reference.

Code:

services:
  jellyfin:
    image: jellyfin/jellyfin:unstable
    container_name: jellyfin-unstable
    user: 105:104
    group_add:
      - "993" # Intel render group
    network_mode: 'host'
    volumes:
      - ./jellyfin-data/config:/config
      - ./jellyfin-data/cache:/cache
      - /media/storage2:/media/storage2:ro
      - /media/library:/media/library:ro
    devices:
      - /dev/dri/renderD128:/dev/dri/renderD128
    restart: 'unless-stopped'
    extra_hosts:
      - "host.docker.internal:host-gateway"

Use the command "getent group render" and replace "993" in mine with your system's render group ID.

Setting getent in the docker compose to my computer's value didn't change anything. I am still completely unable to load JF in the browser if I specify my user 1000:1000, and am still able to load JF if this is commented out.

I have also tried to set up a completely fresh instance of JF Docker, using different /config /cache and /media locations, but nothing is working.

I am truly and totally confused. Any other ideas? The only thing that has changed in my server over the past year is that I moved the media files from one ext4-formatted HD to a ZFS pool. Other than that, I am completely certain that my /config and /cache haven't changed at all.

[bitmap]

Can you get logs from the dead instance? I looked at my DC file and it's different, but not by much and I use the LSIO container anywho.

Only differences: user/group are set with ENV variables, I do NOT use host networking, and my ports are listed. If the Jellyfin container is owned by root, who owns the media you're trying to access?

[Femur4889]

Can you get logs from the dead instance? I looked at my DC file and it's different, but not by much and I use the LSIO container anywho.

Only differences: user/group are set with ENV variables, I do NOT use host networking, and my ports are listed. If the Jellyfin container is owned by root, who owns the media you're trying to access?

Getting logs from dead instance? I'm not sure what that means or how to do that. Can you advise?

I've essentially copied my DC file from Jellyfin's official site. It does not list ports but I'm happy to try that. Can you share your DC yml? 

The JF container was owned by root. I just changed it though. I uncommented "User" so now my container is owned by me, 1000:1000. My user is a member of the docker group. My /cache and /config locations were owned by root and my /media was owned by me. 

So now, I've chown'd /cache and /config to belong to me (1000) while /media was already owned by me. Now that I've done that, I'm able to load the JF container in my browser with "User" uncommented, i.e., belong to me. BUT, I am still unable to load any of the media - the "spinning circle" simply spins and spins.

One thing, though - when my JF container was owned by root, it should have still been able to access my /media which was owned by 1000, correct? or no.

[Femur4889]

There is something very wrong going on here.

I have tried several times to simply start up a fresh JF container with /config and /cache in a totally separate location from this original one. This new container always wants me to log in with an EXISTING user, which there should not be. I looked at /config/system.xml and find that the line <IsStartupWizardCompleted> is set to FALSE, implying that JF should be prompting me to create a new user.

Do I really need to nuke this entire installation of Docker? I really don't even see how that would help. Two days now of trying to get JF to work on this system (which it previously did without problems)...

[3216]

There is something very wrong going on here.

I have tried several times to simply start up a fresh JF container with /config and /cache in a totally separate location from this original one. This new container always wants me to log in with an EXISTING user, which there should not be. I looked at /config/system.xml and find that the line <IsStartupWizardCompleted> is set to FALSE, implying that JF should be prompting me to create a new user.

That’s probably a browser issue. Try accessing the new Jellyfin instance with the browser in private/incognito mode, that should let you configure Jellyfin, and once configured you can go back to a normal browser session and log in.